Question & Answers Service

Tech Stack

Table of Contents

• Project Overview
• Performance & Optimizations

Project Overview

Initially implemented service to replace API and query from a Relational Database

Handled & ETL 10 million rows of data into Postgres

Service initially handled 100 Clients per second with a response time of 62ms

Optimized and horizontally scaled service to handle web scale traffic up to 2000 Clients per second with a response time of 98ms

Initial Service Diagram

Integrating Load Balancer Diagram

---

Performance & Optimizations

Requesting to Retrieve Questions Performance

• Before - Utilizing 1 Server Instance & 1 DBMS Instance

  Max RPS: 1600 clients/sec over 1 minute

  Response Time: 1986ms

  Error Rate: 0%

• After - Utilizing a Load Balancer Instance with 2 Server Instances & 1 DBMS Instance

  RPS: 1600 clients/sec over 1 minute

  Response Time: 79ms

  Error Rate: 0%

Before	After

• Max RPS - Utilizing a Load Balancer Instance with 2 Server Instances & 1 DBMS Instance

  RPS: 2000 clients/sec over 1 minute

  Response Time: 144ms

  Error Rate: 0.1%

Max

Requesting To Retrieve Answers Performance

• Before - Utilizing 1 Server Instance & 1 DBMS Instance

  Max RPS: 1500 clients/sec over 1 minute

  Response Time: 1989ms

  Error Rate: 0%

• After - Utilizing a Load Balancer Instance with 2 Server Instances & 1 DBMS Instance

  RPS: 1500 clients/sec over 1 minute

  Response Time: 82ms

  Error Rate: 0.1%

Before	After

• Max RPS - Utilizing a Load Balancer Instance with 2 Server Instances & 1 DBMS Instance

  RPS: 2000 clients/sec over 1 minute

  Response Time: 98ms

  Error Rate: 0.1%

Max

---

Contributor

Jerome Rodriguez